Google changes the game with Gemini, its most capable AI model

Read time: 2.5 minutes

Welcome back, Superhuman

I give up. I’m simply out of adjectives that I can use to describe the absolute pace of game-changing breakthroughs in AI right now. After several delays and reports of troubles behind the scenes, Google just dropped Gemini — a new AI model that can beat OpenAI’s GPT-4 and ChatGPT. Interesting times lie ahead.

And while Google stole the spotlight with Gemini, Meta also had several major new AI announcements, including a new image generator.

TODAY’S MENU

• Google launches a new era of AI with Gemini

• How to access and use Gemini

• Meta makes major new AI announcements

• Research: Google DeepMind discovers 2.2 million new materials

• 5 new AI tools to get more done with less

• AI Generated Images: Star Wars as a dark fantasy

NEWS

Today in AI & Tech

• Chips Ahoy: Google announces powerful new AI chips.

• More Chips: AMD also launches new AI chips and signs Meta and Microsoft as customers, claims chips outperform Nvidia’s in some aspects.

• Safety First: Meta says fully encrypted messages on Messenger and Facebook are rolling out.

• Small Byte: Apple quietly joins the AI party with release of new AI model framework.

• Bad Look: BBC investigation claims ChatGPT’s custom GPTs could be abused by scammers.

NEXT IN AI

A new era of AI begins with Google Gemini, the company’s most powerful AI model

There’s huge news in AI and then there’s HUGE news in AI. And Google caused the AI equivalent of an earthquake yesterday by announcing Gemini, the company’s most capable AI model yet. It comes in 3 sizes:

• Gemini Ultra: Google’s largest and most capable model for highly complex tasks

• Gemini Pro: Best for scaling across a wide range of tasks

• Gemini Nano: Efficient and best for on-device tasks

So what’s the big deal? While many companies have dedicated billions of dollars and thousands of employees to beating OpenAI’s GPT-4, none have come close yet. Google’s announcement changes that. Gemini Ultra is the first AI model to surpass GPT-4 on several key benchmarks.

But how good is it? Google claims that Gemini Ultra’s performance exceeds current state-of-the-art results on 30 out of the 32 widely-used academic benchmarks:

So what can I use it for? Gemini is multimodal, which means that it can process text, code, images, audio and video. You can use it to do a number of things like:

• Analyzing and generating text, code or data

• Understanding images and generating new ones

• Analyzing video and audio to answer your questions

Here’s a video that shows Gemini’s powerful capabilities:

How is this going to change AI? Gemini’s release shows that it’s possible to beat OpenAI’s GPT-4, something that wasn’t obvious until now. We’re also witnessing the development of different ecosystems within AI. It appeared for a while that OpenAI may have an insurmountable lead, but with general models like Gemini, open source models like Meta’s LLaMa and other more specialized models, consumers and businesses can expect more choice and variety.

Google says Gemini Ultra will be released early next year. For now, Gemini Pro will power Bard. Other Google services will also be upgraded with the latest models soon. Pixel 8 Pro will be the first smartphone to get Gemini Nano.

You can read the full official announcement here. You can also read the full technical report here.

TOGETHER WITH AE STUDIO

Hire a world class AI team for 80% less

Trusted by leading startups and Fortune 500 companies

Building an AI product is hard. Engineers who understand AI are expensive and hard to find. And there's no way of telling who's legit and who's not.

That's why companies around the world trust AE Studio. We help you craft and implement the optimal AI solution for your business with our team of world class AI experts from Harvard, Stanford and Princeton.

Our development, design, and data science teams work closely with founders and executives to create custom software and AI solutions that get the job done for a fraction of the cost.

Book a free consultation session now

p.s. mention you came from Superhuman to get an exclusive $10,000 discount on your first project.

AI AT WORK

How to access and use Gemini Pro today

While Gemini Ultra is a few months away, Gemini Pro is now widely available to test and use on Bard, Google’s alternative to ChatGPT.

Simply, go to the Bard website and log in with your Google account. All you have to do now is enter your prompt and generate answers.

If you’re not sure where to start, here are 5 prompts you can try:

1. Understand Scientific Research: “Explain this paper to me in simple words (share link to paper)“

2. Complete or Grade Assignments: “Review the answer to the question and help me understand if it is solved correctly (share image of assignment)“

3. Create Better Websites and Apps: “Analyze this screenshot of my app/website and give suggestions on how I can improve it (share screenshot)“

4. Learn a Foreign Language: “I am a beginner in (insert language). Help me learn the language by having a conversation with me. Correct my answers along the way.“

5. Generate Recipes from Pictures: “Look at this picture and generate a recipe with ingredients and cooking instructions (share picture of food)“

5 AI Tools to Supercharge Your Productivity

D-ID: Create stunning videos featuring talking avatars anytime and anywhere.

Fathom: Discover podcasts at the speed of thought with AI-powered search, transcripts, chapters, clipping, and highlights.

Glean (sponsored): Generative AI-powered work assistant that searches across all your company’s data. Experience responsible AI that scales to your enterprise needs. Get a demo.

RemoveWatermark: Remove watermarks from images in one click, using AI.

Spellar AI: Enhance your speaking skills and boost your confidence using AI-driven speaking assistant.

NEW PRODUCTS

Meta makes major AI announcements, including a new image generator

Meta wasn’t just going to sit around and have Google steal all of the AI spotlight. The company behind Facebook and Instagram had some major announcements of its own to share:

• New Image Generator: It’s called imagine and it’s powered by Emu, Meta’s AI model for images. You can access it here and use it to generate images with simple text prompts.

• Collaborative Images: You can generate an image in chats with friends on Messenger and Instagram and then use the reimagine feature to generate different versions of the picture together.

• New Experiences with Reels: You can now ask Meta AI to pull specific Reels into group chats. For example, you can ask it to find Reels for a city you plan on visiting.

Meta also announced other major features like watermarking for AI-generated images and upgrades to chatbots. You can find the full list of new releases here.

RESEARCH

A Google AI has discovered 2.2m materials unknown to science

Google’s DeepMind created a machine-learning program called GNoME (Graph Networks for Materials Exploration) that revealed 2.2 million previously unknown crystal structures. This technology predicts new chemical structures using existing chemical structure databases.

This study has greatly increased the number of known candidate materials, which has the potential to revolutionize the discovery of new crystals and improve knowledge of crystal formation.

source

AI-GENERATED IMAGES

Star Wars as a dark fantasy

ADVERTISE WITH US

Acquire new customers and drive revenue by partnering with us

Superhuman is the world’s biggest AI newsletter for businesses and professionals with 500,000+ readers working at the world’s leading startups and enterprises. Companies like Amazon, Calendly, and Notion feature their products in Superhuman. Main ads are typically sold out 4 weeks in advance. You can book future ad spots here.  

🧞 Your wish is my command

Reviews of the day

Thanks for reading.

Until next time!

Zain

p.s. if you want to sign up for this newsletter or share it with a friend or colleague, you can find us here